The Avoidance Obstacle Path Planning Of Mobile Robot In Unknown Environment

In the21st century, as the development of robot research goes deeper, robot researchfield has much more development platform. Mobile robot has more significant advantagesin exploring the complicated and dangerous environment. It can take the place of human tothe South Pole polar, the deep sea, and so on. It is a frontier research subjects the futuredevelopment. And the path planning of mobile robot is a core problem of these subjects, itis very necessary to find a path, securely and efficiently.The research object of this thesis is avoidance obstacle path planning of mobile robot,which analysis the research status of avoidance obstacle path planning of mobile robot inrecent years. It is divided into single robot and robot research based on the different aspect,known and unknown environment, traditional method, intelligent optimization method andother methods. Both the advantages and disadvantages of the different algorithms that usedin different situations have been analysis.First, this thesis introduces two kinds of intelligent optimization method, supportvector machine (SVM) and reinforcement learning. Secondly, this thesis puts forward asemi-supervision SVM of path planning algorithm in known environment. Part of knownidentification samples were applied to give labels of unknown identification samples byusing the similarity matrix, the results were sent to support vector machine process. Theparameter C of SVM is got by using the genetic algorithm, then a path was got, through therandom initial of the known obstacle identification experiments to avoid falling intoextreme, this method reduces the time cost bigger compared with other algorithm. And itwas used for the avoidance obstacle path planning, the detailed design and realization isprocessing. This thesis use experiment to verify the feasibility and effectiveness of thealgorithm.